Mark Rosewater, who heads up design of Magic: the Gathering has talked about randomness' role in a
an article and
a podcast (
summarized on Reddit).
Not all of his points really apply to in-game Pokémon, e.g. it doesn't necessarily matter if the player feels they have a chance to win because they can always soft-reset (or ideally be able to forfeit if the writing is truly on the wall). And while I don't hate RNG in principle, I do think it's poorly executed in Pokémon because very often a single 20% or 30% chance can completely ruin a battle—even if you played well for the previous twenty turns—and that's not even
remotely fun!
I have a theory that because Pokémon is double-blind (i.e. you pick your move without knowing what the foe picked) that predicting your opponent could potentially provide a nice source of "randomness".
In Rock-Paper-Scissors the correct strategy is to pick perfectly randomly. But what if you got two points if you win with the last move that you won with? Well now it's less clear what you should do. Obviously you can't
always pick the same move in a row because then your opponent just counters you each turn, but you probably want to pick the two-point move more than 33% of the time because it's twice as good as any other option if your opponent picks randomly. Except that your opponent knows that you should prefer that move, and also your opponent has their own two-point move, and you know what it is. So lots of mind-games to complicate things.
I think Pokémon is conceptually very close to that RPS variant. On any given turn there's maybe 1–3 valid attacks or switches, and which one you pick depends on all sorts of factors. But unlike in the RPS there are turns where it's pretty obvious what you should do, and I think those turns are pretty important because they provide a nice lull for the player to feel good (or bad) about how things have gone since their last prediction. Unfortunately power creep is ruining this mechanic! IMO it's important that your decisions aren't just coin-flips and that you have a chance to recover from a misprediction, but when attacks are OHKOing and there are lots of mechanics that can snowball (looking at you Soul Heart/Beast Boost) then one slip-up can be just as costly as an ill-timed "fully paralyzed" and it's almost as un-fun.
I don't have a solution fully thought out, but I think I'd approach "fixing" the situation like this:
- Remove some of the more powerful effects. This probably means mega evolution and Z moves need reworking.
- Replace random secondary effects with another mechanic. Probably I'd make Thunderbolt unable to paralyze, and have Discharge paralyze 100% of the time but only on the second consecutive hit.
- Replace accuracy with other mechanics. For things like with at least 80% accuracy I'd probably make them 100% and maybe add an appropriate drawback (e.g. stat drops), and for the others maybe Focus Punch and Sucker Punch can inspire novel solutions to "sometimes this move doesn't work".
- Make each move and ability only able to apply its stat change once (per being switched-in), e.g. the second Swords Dance does nothing,. This should make it less punishing to stay in on a Pokémon that can set up on you.
Of course none of this matters unless the AI is able to make sensible decisions, but that's a topic for another time... :)
---
Sorry for the rambling, but writing this stuff down brought all sorts of ideas that I'd never thought of before and I figured I should get them down.
BTW If anyone has similar feelings about randomness and/or predictions I'd love for you to send me a PM and we can see if there's something we could brainstorm together or collaborate on.
